Abstract
When individual visual objects are moving independently, the performance of human observers is well characterized, even when objects disappear from sight. How ensembles comprised of independent objects moving with a shared goal—such as flocks of birds, schools of fish, or swarms of robots—are visually tracked is less well understood. Here we examined how ensembles of triangles were visually tracked as they moved across a display and how that motion was extrapolated as they passed behind an occluder. In separate experimental blocks, ensembles either moved in lockstep, with each unit translating and rotating identically to the others (cf. a marching band), or with one unit acting as a leader and the others seeking to maintain a given distance behind. Additionally, the overall pattern of motion could either be a parabola or a sinusoid, which varied randomly trial-by-trial. On each trial, the ensemble moved visibly across two thirds of the screen and passed behind an occluder for the remaining third. Participants indicated with mouse clicks the time and vertical location at which the center of the ensemble reached the goal line marked on the occluder. Eye positions was recorded throughout the trial. Participants were more accurate at judging the time of arrival and position when the items moved in lockstep than when the items followed the leader. The pattern of motion did not affect the time of arrival judgement, but the final position was judged more accurately in the sinusoidal motion condition than the parabolic motion condition. These results provide a first look at how ensemble motion is processed and predicted, and suggests that the more that an ensemble acts like a single object, the easier it is to predict.